Scribd
Upload
12 views5 pages

Lesson21 Notes

The document discusses inverse dynamics control for robotics. It explains that inverse dynamics control involves using a model of robot dynamics to compute the necessary torques or joint accelerations to achieve a desired motion. It provides examples of inverse dynamics control in joint space by commanding joint accelerations and in task space by tracking a desired trajectory.

Uploaded by

suianwwww
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
12 views5 pages

Lesson21 Notes

The document discusses inverse dynamics control for robotics. It explains that inverse dynamics control involves using a model of robot dynamics to compute the necessary torques or joint accelerations to achieve a desired motion. It provides examples of inverse dynamics control in joint space by commanding joint accelerations and in task space by tracking a desired trajectory.

Uploaded by

suianwwww
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

control cont'd situ 9.

2
II Inverse Dynamics control IDC
A Joint space IDC SHV 9.3.1

B Task Space IDC SHV 9.3.2

I PD Joint Control

i
Mg if C
q g of Glg u

What happens if we include gravity i e G q to


New eq.pt x̅ sit GCq Kpq
to pointwherecontroltorques
responds
balance the gravitational torques resulting
in steady state error

Can reduce steady state error by increasing Kp but can never

eliminate it w PD control

Also possible to apply Lyapunov analysis to the to


eq.pt
prove stability
Modified PD control w
gravity compensation term
o
U Kpg Kag Glg 8
t as

What happens if we have flexible joints assume no gravity

Consider EIR as robot joint config


q
and IR as vector of motor shaft angles i e
qz
1R and ER
q
Note that is vector of elastic joint deflections
q q
E 8 hn q.is

gfettf.me
e ki
PE PXq.FI q
where 5 0 is
q tK q qe

1 11
i
diag Is
no is
in s

E L M C G
g if q g q k
q q 0 links

K
Jg
motors
q2 q u

of
2n equations
motion
Define PD control law
Let kpqz ka.ge where
u
q 92 qa
for const set point qd ER
a i e using motorfeedback
collocated

Note Collocated feedback is more stabilizing than


collocated feedback
non
joint angle feedback
If G 0 then
q
ÉJq q 95 9 82
e qtMCqq

erectroff
SHU Prob 9 3 0,0
q qz g qa qa
is G A S
by Lasalle
because VG is radiallyunbounded

Not
easy with gravity need to use feedback linearization

II Inverse Dynamics Control


Special case of FB linearization w full actuation
Model based need to identify model first
Useful for dynamic behaviors like trajectory tracking or
energy shaping
Joint space or task space

A Joint Space IDC

q't Clq g if Glq


u i
Mcg
What if we wanted to command joint accelerations
t
g a

Invert the torque to acceleration dynamics


It If ui m.gg
fytftw
inner loop torque controller
decoupled joint dynamics fato
q a

Now we can define an outer loop acceleration


desired joint trajectory
to track
controller a
qᵈ t
by inserting linear dynamics into the double
integrator
is twice differentiable
Assume that t
qᵈ
Outer loop accel controller

a
glt Kp ka

Feedforward Ledack
a
qᵈ t Kp q q t Ka q q t

Kaqi Kp q 0

Closed loop error dynamics are

linear hence FB linearization

Will be stable by choosing Kp70 Kd 0

qg 0,0 is G E s q
t q't
t as
tinal control law

u M q
FÉÉkaj ccq.gl Gq

t.fi
IEiiigIf
Outer loop couldalternatively insert desired
dynamics instead of tracking a reference
trajectory

Ag Ñ q5 v
Cq g q Gq
where Ñ ER I C IR I EIR are

desired dynamics terms

and v is new
input in the shaped

dynamics

a
Mq Eqt G v

This is called energy shaping

You might also like